Comparing BERT-based Reward Functions for Deep Reinforcement Learning in Machine Translation.
Yuki NakataniTomoyuki KajiwaraTakashi NinomiyaPublished in: WAT@COLING (2022)
Keyphrases
- machine translation
- reward function
- reinforcement learning
- reinforcement learning algorithms
- policy search
- markov decision processes
- optimal policy
- state space
- markov decision process
- natural language processing
- information extraction
- target language
- inverse reinforcement learning
- cross lingual
- language independent
- cross language information retrieval
- chinese english
- state variables
- multiple agents
- machine learning
- brazilian portuguese
- natural language
- language resources
- machine translation system
- dynamic programming
- learning algorithm
- transition probabilities
- statistical machine translation
- word alignment
- multilingual documents
- temporal difference
- higher order
- source language
- data mining